Persistent depressive disorder across the adult lifespan: results from clinical and population-based surveys in Germany - BMC ...
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
Nübel et al. BMC Psychiatry (2020) 20:58
https://doi.org/10.1186/s12888-020-2460-5
RESEARCH ARTICLE Open Access
Persistent depressive disorder across the
adult lifespan: results from clinical and
population-based surveys in Germany
Julia Nübel1, Anne Guhn2, Susanne Müllender1, Hong Duyen Le1, Caroline Cohrdes1* and Stephan Köhler2
Abstract
Background: Although the individual and economic disease burden of depression is particularly high for long-term
symptoms, little is known of the lifetime course of chronic depression. Most evidence derives from clinical samples,
and the diagnostic distinction between persistent depressive disorder (PDD) and non-chronic major depression
(NCMDD) is still debated. Thus, we examined characteristics of PDD among clinical vs. non-clinical cases, and the
associated disease burden at a population level.
Methods: Data were drawn from the mental health module of the German Health Interview and Examination
Survey for Adults (DEGS1-MH, 2009–2012, n = 4483) and a clinical sample of PDD inpatients at Charité – Universitätsmedizin
Berlin (2018–2019, n = 45). The DSM-5 definition of PDD was operationalized a priori to the study using interview-based
DSM-IV diagnoses of dysthymia and major depression lasting at least 2 years in both surveys. Additional depression
characteristics (depression onset, self-classified course, suicidality, comorbid mental disorders, treatment history and current
depressive symptoms [Patient Health Questionnaire-9]) were assessed. In the DEGS1-MH, health-related quality of life (Short
Form Health Survey-36, SF-36), chronic somatic conditions, number of sick days (past 12 months) or days with limitations in
normal daily life activities (past 4 weeks), and health service utilization (past 12 months) were compared for PDD vs. NCMDD.
Results: PDD cases from the clinical sample had a significantly earlier depression onset, a higher proportion of self-
classification as persistent course, and treatment resistance than PDD and NCMDD cases in DEGS1-MH. At a population
level, PDD cases showed worse outcomes compared with NCMDD cases in terms of somatic comorbidity, SF-36 mental
component score, and activity limitations owing to mental health problems, as well as a higher risk for outpatient mental
health care contact.
Conclusions: The distinction between PDD and NCMDD proposed for DSM-5 seems warranted. Early onset depression, self-
classification as persistent depressive course, and treatment resistance are suggested as markers of more severe and chronic
depression courses. At a population level, PDD is associated with remarkably higher individual and economic disease
burden than NCMDD, highlighting the need to improve medical recognition of chronic courses and establish specific
treatment concepts for chronic depression.
Keywords: Chronic depression course, Persistent depressive disorder, Prevalence, Disease burden, Health-related quality of
life, General population, Germany, Epidemiology, Cross-sectional studies, Clinical studies
* Correspondence: CohrdesC@rki.de
1
Department of Epidemiology and Health Monitoring, Unit 26 Mental Health,
Robert Koch Institute, PO Box 650261, 13302 Berlin, Germany
Full list of author information is available at the end of the article
© The Author(s). 2020 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0
International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and
reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to
the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver
(http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.Nübel et al. BMC Psychiatry (2020) 20:58 Page 2 of 13 Background does not consider additional lifetime information [25]. More than 300 million people globally were affected by de- Thus, little is known about chronic depression during the pression in 2015, reflecting an increase of about 18% since lifespan (e.g., regarding early vs. late onset depression). 2005 in clinical settings [1]. In terms of years lived with dis- Furthermore, the DSM-5 PDD diagnosis relies predomin- ability, depressive disorder is now a leading contributor to antly on clinical data and the concept of PDD has been non-fatal health loss [2]. Owing to its individual and eco- criticized [30], as its reliability has not been formally ex- nomic disease burden, depression has become a global core amined [31]. However, some researchers still argue for a health challenge of the twenty-first century [3–7]. Social in- diagnostic distinction between chronic and non-chronic surance agencies in Germany have reported an increase in forms of MDD [32]. the frequency of depression and growing health care costs In this study, we aimed to comparatively analyze and dif- owing to working days lost, early retirement, and increased ferentiate characteristics of PDD vs. non-chronic depression health service provision [8, 9]. courses during the lifetime using population-based data However, there are individual differences in depression from the German health monitoring program at the Robert course (i.e., single episodes vs. recurrent episodes), type, Koch Institute and a clinical sample from Charité – Univer- and severity. The enormous economic impact of depres- sitätsmedizin Berlin. We hoped to extend the knowledge of sion on the general population seems particularly related chronic depression beyond the data from clinical samples, to its duration (i.e., long-term), rather than to its severity provide frequency information at a population level, and [10–14]. Primary data indicate that up to 30% of depres- quantify the individual and economic disease burden of sion cases have a chronic course with symptoms that last chronic depression for the general population in Germany. for at least 2 years [12, 15–17]. The 12-month preva- The findings from clinical studies suggest that both the lence of chronic depression is 1.5% [18] and its lifetime indirect costs (e.g., to health-related quality of life or sick prevalence is 3 to 6% [16–18]. In Germany, there is a (leave) days) and the direct costs of health service utilization lack of population-based information on chronic vs. and treatment resistance are much higher for PDD cases non-chronic depression courses. However, secondary than for non-chronic cases. data from national health insurance companies indicate The study objectives were 1) the classification of that up to two-thirds of medical depression diagnoses chronic vs. non-chronic depression courses at a popula- take a chronic course over at least 2 years (repeated tion level, 2) the identification of PDD characteristics in registration irrespective of type or severity) [19]. a clinical vs. population-based sample, and 3) the com- Furthermore, chronic depression may have an earlier parison of PDD vs. non-chronic MDD (NCMDD) in onset (before 21 years of age) [14, 20–22] and worse terms of associations with health-related correlates at a outcomes than non-chronic depression, such as single or population level. recurrent depressive episodes with full inter-episode recov- ery. Chronic depression is characterized by higher comor- Methods bidity rates [12–15, 18, 20, 22], somatic morbidity [14, 15], Data basis and depression assessment suicidality [14, 20, 22], reduced somatic and psychological Data for the nationwide representative analyses were well-being and health-related quality of life [12–14, 23], drawn from the first wave of the German Health Interview lower employment rates [24], longer delays for treatment and Examination Survey for Adults (DEGS1, field work [15], and limited effects of psychotherapeutic or psycho- 2008–2011, n = 7115) and its mental health module pharmacological treatment [10, 11, 13, 25–27], all of which (DEGS1-MH, field work 2009–2012, n = 4483), which in- indicate its enormous direct and indirect costs. cluded 18- to 79-year-old participants from statutory as However, comparisons of the characteristics, prevalence, well as private health insurances based on a two-stage and disease burden of chronic vs. non-chronic depression clustered random sampling procedure (step 1: random is hampered by two facts: most knowledge derives from sampling of study locations from all municipal communi- clinical samples [15] and prevalence estimates differ, be- ties; step 2: random sampling of participants from the cause a generally accepted definition of chronic depression population-registries in each sampled study location). The was lacking until the American Psychiatric Association in design and methods are described in detail elsewhere [33– 2013 decided to include a new depressive subtype, persist- 35]. DEGS1 and DEGS1-MH were part of the German ent depressive disorder (PDD), in the latest version of the health monitoring program and provided data about the Diagnostic and Statistical Manual of Mental Disorders, health of the non-institutionalized population in Germany Fifth Edition (DSM-5) [28, 29]. PDD is defined as depres- based on self-rated questionnaires and a standardized sion that persists for at least 2 years. The PDD subtype is computer-assisted Interview conducted by study physi- thus a combination of the DSM-IV diagnoses of (lasting) cians (CAPI). Mental disorders, including MDD and DD, major depressive disorder (MDD) and dysthymic disorder were assessed by trained interviewers based on the World (DD). However, even the new PDD diagnostic category Health Organization Composite International Diagnostic
Nübel et al. BMC Psychiatry (2020) 20:58 Page 3 of 13
Interview (CIDI). The CIDI is a standardized fully struc- a general medical condition (criterion D) or attributable to
tured computer-assisted clinical face-to-face interview and grief (criterion E). DD diagnosis requires depressed mood
is an internationally established measure of mental disor- for most of the day and for at least 2 years (criterion A),
ders [36–38]. A modified German version of the CIDI was and at least two out of six depression symptoms (criterion
used in DEGS1-MH [33] to assess mental disorders ac- B). During the 2 years, the total recovery time should not
cording to the DSM-IV-TR diagnostic criteria [39]. The have exceeded more than 2 months (criterion C) and the
CIDI provides lifetime information about symptoms (e.g., symptoms should have caused clinically significant distress
age of onset, recurrence and duration of episodes) that or impairment (criterion H). Exclusion criteria include
permits analysis of the course of depression over the life- manic/hypomanic episodes (criterion E), symptoms owing
span. After participants with missing information on to the direct physiological effects of a substance or a gen-
affective disorders were excluded (n = 75), the final study eral medical condition (criterion G), or symptoms occur-
sample was n = 4408. ring during the course of a psychotic disorder (criterion
Data were also obtained from a clinical sample re- F). Furthermore, DSM-IV DD diagnosis requires the ab-
cruited at the Charité – Universitätsmedizin Berlin (n = sence of a major depressive episode during the first 2 years
60). Patients with a professional diagnosis of PDD ac- of occurrence (criterion D). However, DSM-5 no longer
cording to DSM-5 [28] were treated for 12 weeks with a includes this criterion for PDD diagnosis, and MDD cri-
specialized chronic depression intervention: the Cogni- teria may be continuously present for 2 years.
tive Behavioral Analysis System of Psychotherapy Thus, subjects with lifetime or 12-month MDD ac-
(CBASP; [27, 40]). Patients were either directly referred cording to CIDI or SCID I who also report a lifetime
from the outpatient clinic of the Charité, from inpatient maximum episode duration of at least 104 weeks, as well
wards of other hospitals from all parts of Germany, or as subjects (concurrently) fulfilling the DD diagnostic
from outpatient psychiatrists. Treatment was reimbursed criteria (irrespective of DSM-IV criterion D), were classi-
by statutory health insurances. Exclusion criteria for the fied as lifetime PDD cases. The remaining MDD cases
inpatient CBASP were a history of psychotic episodes, were categorized as non-chronic cases (NCMDD). The
bipolar I or II disorders, comorbid substance depend- grouping of PDD and NCMDD was carried out a priori
ence with less than 3 months of abstinence, severe forms to the study. Cases with missing responses for maximum
of autism, and organic mental disorders. All patients episode duration and missing information on diagnostic
treated at the ward from 2013 to 2018 were invited for a criteria of DD have been omitted. In the clinical sample,
subsequent follow-up interview for the purpose of the health professional-diagnosed PDD was validated via
present study. These interviews were conducted from SCID I for all patients.
October 2018 to March 2019 to collect lifetime informa-
tion on course and type of depression and comorbid Depression characteristics
mental disorders using the Structured Clinical Interview Age of depression onset and the number of depressive
for DSM-IV (SCID I; [41]) and self-rated questionnaires. episodes were assessed in both diagnostic interviews.
To allow comparison with the epidemiological sample, History of suicidality was also assessed in both surveys
additional questions based on the CIDI depression sec- based on CIDI questions about thoughts of death or sui-
tion were included. The final clinical study sample com- cide, suicide plans, or attempted suicide.
prised n = 45 patients, aged 24–66 years. Subjects of DEGS1-MH and patients of the clinical
sample rated their course of depression based on CIDI
Definition and operationalization of (non-)chronic depression section diagram on the following categories:
depression single episode (remitted), single episode (acute), recur-
For this study, the definition of chronic depression was ring episodes, single episode with chronic course, per-
based on the DSM-5 PDD diagnosis and drawn from the sistent depressive course, double depression, or other.
DSM-IV-based diagnoses of MDD or DD derived from MDD symptoms according to DSM-IV were assessed
the SCID I or CIDI. According to DSM-IV, MDD diagno- using the German version of the internationally established
sis requires the persistence of at least five out of nine Patient Health Questionnaire (PHQ-9). The PHQ-9 con-
depressive symptoms on nearly every day for 2 weeks or sists of nine items assessing the presence and frequency of
longer, of which at least one is depressed mood or depressive symptoms during the past 2 weeks. Summed
decreased interest/pleasure (criterion A). Furthermore, scores ≥10 indicate current depressive symptoms [42, 43].
clinically significant distress and impairment associated The number of comorbid mental disorders (lifetime)
with these symptoms are necessary (criterion C). MDD was categorized as none, one, and at least two of the
exclusion criteria include lifetime manic/hypomanic epi- CIDI- or SCID I-based diagnoses of mental disorders
sodes (criterion B) and depressive symptoms solely attrib- during the lifetime. As some mental disorders were in-
utable to the direct physiological effects of a substance or cluded in the exclusion criteria for the clinical sample,Nübel et al. BMC Psychiatry (2020) 20:58 Page 4 of 13
the following comorbid diagnoses were assessed: panic at the time of the clinical follow-up as well as of the
disorder, agoraphobia, generalized anxiety disorder, social DEGS1 mental health module assessment and catego-
phobia, specific phobias, obsessive–compulsive disorder, rized into age groups (18–34, 35–49, 50–64, and 65–79
posttraumatic stress disorder, pain and somatoform disor- years). Marital status was dichotomized into married
ders, substance abuse and dependence (excluding nico- and living with partner vs. married and not living with
tine), anorexia nervosa, bulimia nervosa, and binge eating partner/single/never been married/divorced/widowed.
disorder. The Comparative Analysis of Social Mobility in Indus-
Self-reported mental health treatment during the life- trial Nations (CASMIN) scale was used to classify
time was assessed based on the CIDI questions in both responses on educational level into low, medium, and
DEGS1-MH and the clinical sample. The number of high. In DEGS1-MH, structural social support was
antidepressant treatments and the number of psycho- assessed using the Oslo-3 Social Support Scale [48].
therapies were each categorized as none, one, and at
least two treatments. Treatment resistance was defined
for cases with at least two reported antidepressant treat- Statistical analysis
ments, approaching the definition of Thase and Rush Frequency and mean estimates of the sample characteris-
(medication resistance to two or more adequate trials of tics are reported with 95% confidence intervals (95% CI).
antidepressants) [44]. At a population level, prevalence estimates for lifetime
MDD and DD are reported. Conditional frequencies for
Health-related correlates chronic vs. non-chronic courses among lifetime MDD cases
Several health-related correlates were assessed in DEGS1- are reported. Prevalence estimates for PDD and NCMDD
MH: self-rated health (dichotomized into fair/poor vs. could not be provided owing to many missing responses
good/very good/excellent) and health-related quality of life for self-reported maximum episode duration, resulting in a
(past 4 weeks) were assessed using the German version of high proportion of MDD with unknown chronicity.
the Short Form Health Survey-36 (SF-36) version 2 [45, Frequency and mean estimates for depression character-
46]). The physical component score (PCS) and the mental istics are reported with 95% CI for PDD cases in the clin-
component score (MCS) were used as total scales with a ical sample and for PDD and NCMDD cases in the
mean value of 50 and a standard deviation of 10 (higher population-based sample. The significance (p < .01) of dif-
values indicate better health-related quality of life). The ferences between the clinical sample and the DEGS1-MH
number of days with limitations in normal daily life activ- sample was indicated by non-overlapping 95% Cis [47]
ities owing to physical vs. mental health problems (includ- and sizes of significant effects for independent groups with
ing limitations owing to substance use) during the past 4 different sample size are indicated by Cohen’s d (small =
weeks were also assessed [see 23]. The self-reported num- 0.2, medium = 0.5, large = 0.8). Statistical significance of
ber of sick days during the past 12 months was assessed in differences between PDD and NCMDD characteristics in
DEGS1 (irrespective of occupational status), as well as DEGS1-MH were evaluated using the Rao–Scott chi-
self-reported information on health service use during the square test for categorical variables, and the Wilcoxon–
past 12 months (number of outpatient physician visits, Mann–Whitney test for continuous variables, using a two-
outpatient psychiatric or psychotherapeutic contacts, and sided significance level of 0.05.
number of nights in hospital). The number of chronic Health-related correlates are shown for DEGS1-MH
somatic conditions reported in DEGS1 was classified as PDD vs. NCMDD cases with 95% CI, to enable the com-
none, one, and at least two of the following somatic condi- parison of the associated individual and economic disease
tions [see 47]: myocardial infarction (lifetime), chronic burden at a population level. Effect estimates for health-
heart failure (lifetime), stroke (lifetime), osteoarthritis (life- related correlates in cases with PDD vs. NCMDD were
time), rheumatoid arthritis (past 12 months), osteoporosis based on logistic, linear, negative binomial, or zero-inflated
(lifetime), gout (past 12 months), bronchial asthma (past negative binomial regression models, including health-
12 months), cirrhosis of the liver (lifetime), hepatitis (past related correlates as dependent variables and depression
12 months), gastric-duodenal ulcer (past 12 months), can- course (PDD vs. NCMDD) as the independent variable (ref-
cer (lifetime), Parkinson’s disease (lifetime), epilepsy (past erence: NCMDD). All analyses were adjusted for sex, age
12 months), hypertension (past 12 months), dyslipidemia group, educational level, marital status, social support,
(past 12 months), renal failure (lifetime), and inflammatory chronic somatic conditions (except for analysis of the num-
bowel disease (past 12 months). ber of chronic somatic conditions as an outcome variable),
and PCS (except for analysis of PCS as an outcome
Other measures variable) [see 49]. The results of the unadjusted regression
Sociodemographic variables included sex, age, marital analyses are included as supplementary data (see Add-
status, and educational level. Age was assessed in years itional file 1) and only described if divergent. StatisticalNübel et al. BMC Psychiatry (2020) 20:58 Page 5 of 13
significance was evaluated based on a two-sided significance 39.2%) and demonstrated a significantly higher educa-
level of 0.05. tional level than the DEGS1-MH sample (as indicated by
All statistical analyses were performed using Stata 15.1 non-overlapping 95% Cis).
(StataCorp, College Station, Texas, USA). For DEGS1-MH,
all analyses were performed using the Stata survey design Chronic depression at a population level
procedures to account for clustering and weighting of the Among cases with lifetime MDD in DEGS1-MH (14.5%),
study sample. Thus, survey-specific weighting factors were 18.2% reported a maximum episode duration of at least 2
used to adjust the sample to the demographic distribution years, and 15.4% fulfilled the diagnostic criteria of concur-
of the population in Germany as on 31st December, 2010, rent DD (without considering criterion D). Overall, 36.5%
regarding sex, age, educational status, federal state, nation- of cases with a lifetime CIDI diagnosis of MDD were clas-
ality, and the probability of participation in the mental sified as chronic MDD cases; the remaining 63.5% were
health module subsequent to the core survey [33, 50]. categorized as NCMDD cases. In addition to chronic
In addition, we calculated post-hoc power analyses to MDD, PDD also comprised subjects with solely lifetime
test for appropriate test power based on the present sam- DD (1.3%, without considering criterion D).
ple sizes.
Characteristics of chronic depression in a clinical sample
Results and at a population level
Sample characteristics PDD cases from the clinical sample had a significantly
Sample characteristics of the clinical sample and earlier disease onset than cases with PDD and NCMDD
DEGS1-MH sample are shown in Table 1. The DEGS1- in DEGS1-MH (age of disorder onset < 21 years: 73.3%
MH sample was comparable to the clinical sample on vs. 24.7% vs. 32.2%, see Table 2). Suicidality (thoughts of
age and sex, except for the proportion of participants death/suicide, or having suicide plans/attempts) was re-
aged 50–64 years (higher in the clinical sample) and 65– ported more often by PDD cases in the clinical sample
79 years (higher in DEGS1-MH participants). Clinical than by PDD or NCMDD cases in the DEGS1-MH sam-
sample patients more frequently lived alone (88.9% vs. ple (95.5% vs. 86.4% vs. 86.2%), as was attempted suicide
Table 1 Sample characteristics of the clinical and the population-based sample
Clinical sample1 DEGS1-MH2 Cohen’s d
(n = 45) (n = 4408)
Sex, % (95% CI)
Male 53.3 (37.8–66.7) 49.1 (47.1–51.0) 0.84
Female 46.7 (33.3–62.2) 50.9 (49.0–52.9) 0.84
Age, mean (95% CI) 47.1 (43.6–50.6) 48.0 (47.4–48.6) 0.18
Age group (years), % (95% CI)
18–34 22.2 (12.1–37.2) 24.8 (23.1–26.5) 0.61
35–49 24.4 (13.8–39.6) 29.2 (27.5–30.9) 1.07
50–64 51.1 (36.3–65.8) 25.4 (24.0–26.9) 5.83
65–79 2.2 (0.3–15.1) 20.7 (19.4–22.0) 4.53
Marital status3, % (95% CI)
Married and living with partner 11.1 (2.2–20.0) 60.8 (58.5–63.2) 26.87
Married and not living with partner/Single/Divorced/Widowed 88.9 (80.0–97.8) 39.2 (36.9–41.6) 10.18
4
Educational level , % (95% CI)
Low 2.3 (0.0–6.8) 35.0 (32.6–37.6) 6.90
Medium 36.4 (22.7–52.3) 50.8 (48.7–52.9) 2.89
High 61.4 (47.7–45.0) 14.2 (12.6–16.0) 13.51
CI confidence interval, CIDI Composite International Diagnostic Interview, Cohen’s d effect size computed for groups with different sample size, by adjusting the
calculation of the pooled standard deviation with weights for the sample sizes
1
Clinical Sample at Charité - Universitätsmedizin Berlin: 2018–2019; n = 45 re-participants based on n = 60 patients who participated in inpatient treatment with
the Cognitive Behavioral Analysis System of Psychotherapy (CBASP [40]), age range: 24–66 years
2
German Health Interview and Examination Survey for Adults, mental health module (DEGS1-MH): 2009–2012, weighted for population structure as of 31st
December 2010; age range: 18–79 years; n = 4408 with full CIDI mood disorders section
3
In DEGS1-MH, n = 32 subjects did not provide information on marital status
4
Categorization according to the Comparative Analysis of Social Mobility in Industrial Nations (CASMIN) scale. There were missing values for n = 1 subject in the
clinical sample and n = 17 participants in the DEGS1-MH sampleNübel et al. BMC Psychiatry (2020) 20:58 Page 6 of 13
Table 2 Characteristics of cases with (non-)chronic depression during the lifetime in clinical and population-based samples
Cases with PDD in Cases with PDD Cases with NCMDD p-value5
clinical sample1 in DEGS1-MH2 in DEGS1-MH
(n = 45) (n = 179) (n = 205)
Age of disorder onset < 21 years3, % (95% CI) 73.3 (58.0–84.5) 24.7 (18.2–32.6) 32.2 (23.8–41.9) 0.192
Thoughts of death/suicide, or suicide 95.5 (82.8–98.9) 86.4 (78.7–91.6) 86.2 (77.6–91.8) 0.960
plans/attempts3, % (95% CI)
Attempted suicide3, % (95% CI) 36.4 (23.2–52.0) 16.2 (9.2–26.8) 11.7 (6.9–19.3) 0.393
Self-reported depression course3, % (95% CI) < 0.001
Single episode, remitted – 2.3 (0.9–5.4) 17.7 (10.4–28.3)
Single episode, acute – 5.7 (2.7–11.7) 4.0 (1.6–9.9)
Recurring episodes 2.3 (0.3–15.5) 20.9 (14.4–29.3) 55.1 (45.0–64.8)
Single episode, chronic course 25.0 (14.1–40.4) 24.3 (14.9–36.9) 5.9 (3.1–10.9)
Persistent depressive course 50.0 (35.1–64.9) 24.6 (16.9–34.4) 2.0 (0.5–7.9)
Double depression 15.9 (7.5–30.5) 19.3 (12.4–28.8) 11.9 (6.4–21.3)
Other 6.8 (2.1–19.9) 2.9 (0.9–9.4) 3.5 (1.1–9.9)
No. of episodes3, mean (95% CI) 2.8 (1.9–3.6) 13.7 (8.4–19.0) 7.4 (5.1–9.7) < 0.001
4
No. of comorbid mental disorders (lifetime) , % (95% CI) 0.071
0 46.7 (32.2–61.7) 28.7 (20.8–38.2) 36.9 (28.0–46.7)
1 33.3 (20.8–48.8) 28.9 (21.7–37.5) 37.4 (29.0–46.8)
2 13.3 (5.9–27.3) 31.0 (21.8–42.0) 16.4 (10.5–24.8)
≥3 6.7 (2.1–19.4) 11.4 (6.2–20.1) 9.3 (5.0–16.6)
Current depressive symptoms (PHQ-9 ≥ 10), % (95% CI) 66.7 (51.2–79.2) 44.9 (35.1–55.1) 18.6 (12.1–27.5) < 0.001
No. of psychotherapeutic treatments (lifetime) 0.271
0 – 87.2 (76.3–93.5) 92.6 (86.6–96.0)
1 9.1 (2.8–2.6) 10.1 (4.7–20.3) 6.5 (3.3–12.6)
≥2 90.9 (74.1–97.2) 2.7 (1.1–6.7) 0.9 (0.3–2.4)
No. of antidepressant medications (lifetime) 0.674
0 2.4 (0.3–16.2) 79.5 (71.7–85.6) 75.1 (66.2–82.3)
1 16.7 (7.9–31.8) 11.4 (7.4–17.1) 12.7 (7.8–20.1)
≥2 81.0 (65.6–90.5) 9.1 (5.1–15.9) 12.2 (7.0–20.3)
CI confidence interval, PDD persistent depressive disorder, NCMDD non-chronic major depressive disorder, SCID I Structured Clinical Interview for DSM-IV, CIDI
Composite International Diagnostic Interview
1
Clinical Sample at Charité - Universitätsmedizin Berlin: 2018–2019; n = 45 re-participants based on n = 60 patients who participated in inpatient treatment with
the Cognitive Behavioral Analysis System of Psychotherapy (CBASP [40]), age range: 24–66 years
2
German Health Interview and Examination Survey for Adults, mental health module (DEGS1-MH): 2009–2012, weighted for population structure as of 31st
December 2010; age range: 18–79 years; n = 4408 with full CIDI mood disorders section
3
Based on CIDI
4
Based on SCID I or CIDI diagnoses of panic disorder, agoraphobia, generalized anxiety disorder, social phobia, specific phobias, obsessive–compulsive disorder,
posttraumatic stress disorder, pain and somatoform disorders, substance abuse and dependence (excluding nicotine), anorexia nervosa, bulimia nervosa, and
binge eating disorder
5
p-value based on Rao–Scott chi-square test for categorical variables and based on Wilcoxon–Mann–Whitney test for continuous variables in DEGS1-MH. Bold type
indicates significant differences between subjects with PDD and NCMDD in DEGS1-MH (local significance level α = 0.01)
(36.4% vs. 16.2% vs. 11.7%), but the significance of these dif- 24.3% vs. 5.9%). Furthermore, a significantly higher propor-
ferences remains unclear with one exception: the propor- tion of clinical PDD patients showed a persistent depressive
tion of PDD patients in the clinical sample that attempted course compared with PDD and NCMDD DEGS1-MH
suicide was more than three times greater than the propor- cases (50.0% vs. 24.6% vs. 2.0%), and a smaller frequency of
tion of NCMDD cases in DEGS1-MH. Regarding self- recurring episodes (2.3% vs. 20.9% vs. 55.1%; significant dif-
reported depression course, PDD cases differed significantly ference only for clinical PDD patients compared with
from NCMDD cases in DEGS1-MH (p < 0.001). Both PDD NCMDD cases). Accordingly, cases with PDD in DEGS1-
groups showed significantly higher rates of a chronic course MH reported a significantly higher mean number of epi-
of a single episode compared with NCMDD cases (25.0 and sodes in total (13.7) compared with both clinical PDDNübel et al. BMC Psychiatry (2020) 20:58 Page 7 of 13
patients (2.8) and NCMDD cases in DEGS1-MH (7.4, p < and 4. The risk of experiencing fair or poor self-rated
0.001). Comorbid mental disorders seemed to be more pro- health was significantly higher among PDD cases (36.8%)
nounced among cases with PDD and NCMDD in DEGS1- than among NCMDD cases (20.4%, odds ratio [OR] = 2.0,
MH compared with the clinical sample, but the significance p = 0.041). Mean health-related quality of life (past 4 weeks)
of these differences remains unclear. There was a trend for was lower among PDD cases for PCS (47.1 vs. 50.7, signifi-
higher comorbidity among PDD cases than among cant only for crude effect estimates, see Additional file 1)
NCMDD cases in DEGS1-MH (p = 0.071). The prevalence and MCS (34.5 vs. 43.8, β = − 8.2, p < 0.001). Accordingly,
of current depressive symptoms was highest among the the mean number of days with activity limitations (past 4
clinical PDD patients (PHQ-9 ≥ 10: 66.7%), and significantly weeks) owing to mental health problems was higher for
higher among PDD cases compared with NCMDD cases in PDD than for NCMDD (5.4 vs. 2.4, incidence rate ratio
DEGS1-MH (44.9% vs. 18.6%, p < 0.001). Furthermore, [IRR] = 2.6, p < 0.001). There was also a trend for more re-
clinical PDD cases showed a significantly higher treatment ported limitation days owing to physical health problems
resistance than PDD and NCMDD cases in DEGS1-MH, in for PDD compared with NCMDD cases (5.3 vs. 3.1, IRR =
terms of the proportion of cases reporting at least two psy- 1.4, p = 0.091). There was also a higher risk of sick days
chotherapeutic treatments (90.9% vs. 2.7% vs. 0.9%) or anti- during the past 12 months for PDD cases (34.2 vs. 14.8),
depressant medications (81.0% vs. 9.1% vs. 12.2%) during but this was only significant in the unadjusted analysis (see
the lifetime. Most PDD and NCMDD cases in DEGS1-MH Additional file 1). Indicators of health service use during
reported no psychotherapeutic treatment (87.2 and 92.6%) the past 12 months showed higher utilization rates for PDD
or antidepressant medication (79.5 and 75.1%). than for NCMDD cases for the mean number of outpatient
psychiatric or psychotherapeutic contacts (5.7 vs. 1.7, IRR =
Health-related correlates of (non-)chronic depression at a 2.7, p = 0.006). There was also a trend for PDD cases to re-
population level port a higher mean number of nights in hospital compared
The associations of PDD vs. NCMDD with health-related with NCMDD cases (3.9 vs. 0.9, IRR = 1.9, p = 0.065). The
correlates based on DEGS1-MH are shown in Tables 3 mean number of outpatient physician visits (4.3 vs. 3.6) was
Table 3 Health-related correlates in cases with PDD vs. NCMDD during the lifetime at a population level1
PDD NCMDD
(n = 179) (n = 205)
Fair/poor self-rated health, % (95% CI) 36.8 (28.8–45.6) 20.4 (13.8–29.0)
Health-related quality of life (past 4 weeks), mean (95% CI)
Physical component score 47.1 (44.9–49.4) 50.7 (48.8–52.7)
Mental component score 34.5 (32.2–36.8) 43.8 (41.7–45.9)
No. of days with activity limitations (past 4 weeks), mean (95% CI)
Owing to mental health problems 5.4 (4.1–6.7) 2.4 (1.2–3.6)
Owing to physical health problems 5.3 (3.9–6.7) 3.1 (1.9–4.2)
No. of sick days (past 12 months), mean (95% CI) 34.2 (17.6–50.8) 14.8 (9.9–19.7)
No. of outpatient physician visits (past 12 months), mean (95% CI) 4.3 (3.4–5.2) 3.6 (2.9–4.2)
No. of outpatient psychiatric or psychotherapeutic contacts 5.7 (2.1–9.4) 1.7 (0.7–2.7)
(past 12 months), mean (95% CI)
No. of hospital nights (past 12 months), mean (95% CI) 3.9 (1.2–6.6) 0.9 (0.5–1.3)
No. of chronic somatic conditions2, % (95% CI)
0 42.8 (33.7–52.5) 63.9 (53.7–72.9)
1 31.0 (22.9–40.4) 20.6 (13.6–30.0)
2+ 26.2 (19.6–34.1) 15.6 (10.0–23.5)
CI confidence interval, PDD persistent depressive disorder, NCMDD non-chronic major depressive disorder, CIDI Composite International Diagnostic Interview
1
German Health Interview and Examination Survey for Adults, mental health module (DEGS1-MH): 2009–2012, weighted for population structure as of 31st
December 2010; age range: 18–79 years; n = 4408 with full CIDI mood disorders section
2
Myocardial infarction (lifetime), chronic heart failure (lifetime), stroke (lifetime), osteoarthritis (lifetime), rheumatoid arthritis (past 12 months), osteoporosis
(lifetime), gout (past 12 months), bronchial asthma (past 12 months), cirrhosis of the liver (lifetime), hepatitis (past 12 months), gastric-duodenal ulcer (past 12
months), cancer (lifetime), Parkinson’s disease (lifetime), epilepsy (past 12 months), hypertension (past 12 months), dyslipidemia (past 12 months), renal failure
(lifetime), and inflammatory bowel disease (12 months)
Bold type indicates significant associations between depression course (PDD vs. NCMDD) and health-related correlates (at local significance level α = 0.05, resulting
from multiple (continuous outcome), multiple negative binomial (dichotomous outcome) or multinomial (multinomial outcome) regression analyses; see Table 4
for detailed statistical parameters)Nübel et al. BMC Psychiatry (2020) 20:58 Page 8 of 13
Table 4 Effect estimates for health-related correlates in cases of PDD vs. NCMDD (ref.) during the lifetime1
Effect estimate p-value
(95% CI)
Fair/poor self-rated health OR 2.0 (1.0–3.9) 0.041
Health-related quality of life (past 4 weeks)
Physical component score β −1.2 (-3.9 – -1.5) 0.374
Mental component score β −8.2 (-11.5 – -4.9) < 0.001
No. of days with activity limitations (past 4 weeks)
Owing to mental health problems IRR 2.6 (1.6–4.3) < 0.001
Owing to physical health problems IRR 1.4 (0.9–2.1) 0.091
No. of sick days (past 12 months) IRR 1.3 (0.9–1.8) 0.193
No. of outpatient physician contacts (past 12 months) IRR 1.0 (0.8–1.3) 0.847
No. of outpatient psychiatric/psychotherapeutic contacts IRR 2.7 (1.3–5.4) 0.006
(past 12 months)
No. of hospital nights (past 12 months) IRR 1.9 (1.0–3.8) 0.065
No. of chronic somatic conditions
0 ref.
1 RRR 2.8 (1.3–5.8) 0.008
2+ RRR 3.2 (1.4–7.0) 0.004
Regression models include health-related correlates as dependent variables and depression course (PDD vs. NCMDD) as the independent variable (reference:
NCMDD). All analyses were adjusted for sex, age group, educational level, marital status, social support, chronic somatic conditions (except for analysis of the no.
of chronic somatic conditions as an outcome variable), and PCS (except for analysis of PCS as an outcome variable). OR: Odds ratio from logistic regression; β: β
coefficient from linear model; IRR: incidence rate ratio from negative binomial regression or zero-inflated negative binomial regression; RRR: relative risk ratio from
multinomial logistic regression; p-value for testing the effect of depression course (test for OR/IRR/RRR = 1 or β = 0)
CI confidence interval, PDD persistent depressive disorder, NCMDD non-chronic major depressive disorder, PCS physical component score, CIDI Composite
International Diagnostic Interview
1
German Health Interview and Examination Survey for Adults, mental health module (DEGS1-MH): 2009–2012, weighted for population structure as of 12/31/2010;
age range: 18–79; n = 4408 with full CIDI mood disorders section
Bold type indicates significant associations between depression course (PDD vs. NCMDD) and health-related correlates (at local significance level α = 0.05, resulting
from multiple (continuous outcome), multiple negative binomial (dichotomous outcome) or multinomial (multinomial outcome) regression analyses)
only significantly higher for PDD cases in the unadjusted in a population-based sample, with an error probability of
analysis (see Additional file 1). Furthermore, somatic co- α = 0.05 and at the power level of 1.00 (see Table 4).
morbidity was significantly higher for PDD vs. non-chronic
cases. The risk of having one chronic condition (31.0% vs. Discussion
20.6%, relative risk ratio [RRR] = 2.8, p = 0.008) or at least Based on a nationally representative sample of the general
two comorbid conditions (26.2% vs. 15.6%, RRR = 3.2, p = adult population in Germany, more than one-third (36.5%)
0.004) was approximately 3-fold for PDD. In contrast, most of all subjects fulfilling MDD criteria showed a chronic de-
NCMDD cases (63.9%) had no somatic comorbidity at all pression course with maximum episode duration of at least
(vs. 42.8% of PDD cases). 2 years and/or concurrent dysthymia at least once during
the lifetime. This rate is slightly higher than previous inter-
Post-hoc power analyses national frequency estimates, which reported a chronic
Results from post-hoc power analyses with the help of course for only 21 to 30% of depressed cases [12, 15–17].
G*Power 3 [51] suggest that that the present sample size This inconsistency can be explained by different definitions
of n = 429 individuals was sufficient for the detection of of chronic depression: previous prevalence based solely on
moderate effects (ω = 0.30) within a chi-square goodness- episode duration, without considering MDD cases with
of-fit test comparing PDD vs. NCMDD in clinical and double depression (i.e., MDD and DD).
population-based samples for each health-related correlate
and an error probability of α = 0.05, at the power level of More severe PDD cases in the health care system
1.00 (see Table 2). Moreover, results from post-hoc power Overall, our estimated frequency for DEGS1-MH cases
calculation suggest that the present sample size of n = 285 with MDD that had a chronic course during the lifetime
individuals was sufficient for the detection of moderate ef- (36.5%) was much lower than the proportion reported
fects (f2 = 0.15) within a multiple regression design con- from national health insurance data (65%) [19]. However,
taining five predictors (PDD vs. NCMDD, age, sex, marital previous findings show that among cases with CIDI-based
status, educational level) on each health-related correlate MDD, 65.4% did not report any health service use forNübel et al. BMC Psychiatry (2020) 20:58 Page 9 of 13
mental health problems [52]; and service use increased only been observed by trend, and our clinical sample of
with depression severity [52]. Thus, particularly severe PDD patients demonstrated even less comorbidity than
(and chronic) depression cases may eventually access the interview-defined cases. This may be related to differ-
health care system, leading to higher proportions of ences in the diagnostic tools (SCID I vs. CIDI). Add-
chronic depression courses based on health insurance data itionally, personality disorders, which account for a large
[see 19] compared with frequency estimates for interview- proportion of comorbidities in the reviews, were not
based MDD cases at a population level. assessed in both samples. However, interpersonal prob-
Consequently, our comparisons of depression character- lems as indicated by the social functioning subscale of
istics indicate that PDD cases in the health care system the SF-36 were significantly reduced among PDD cases
are more severely affected, since clinical sample PDD pa- as compared to NCMDD cases in DEGS1-MH (post hoc
tients showed a pronounced long-term duration owing to sensitivity analysis; PDD: M = 61.47, 95%CI = 55.89–
earlier onset (73.3% vs. 24.7% with age onset ≤21 years) 67.05; NCMDD: M = 76.77, 95%CI = 72.67–80.86). Con-
and significantly higher rates of treatment resistance sequently, only minor and non-significant differences in
(81.0% vs. 9.1% reported at least two antidepressant medi- mental comorbidity have been observed between
cation trials) compared with interview-defined PDD cases interview-defined PDD and NCMDD cases.
at a population level, as well as a higher proportion of self-
classified persistence of depressive course. Furthermore,
Higher disease burden for chronic vs. non-chronic
the prevalence of attempted suicide during lifetime was
depression
higher among clinical PDD patients as compared to the
The comparison of interview-defined cases of PDD vs.
DEGS1-MH PDD cases (but nonsignificant) and more
NCMDD at a population level highlighted that several
than three times higher than among NCMDD cases.
health-related correlates indicate higher individual and
Considering the existing literature, our results are in
economic disease burden for chronic depression courses.
line with clinical findings. For instance, lifetime preva-
On the individual level, there was a remarkably higher
lence of treatment resistance for depression was 81.8%
prevalence of current depressive symptoms (as assessed by
in patients with long-term depression vs. 60.7% in
PHQ-9) among PDD cases than among NCMDD cases, as
patients with depression lasting less than 2 years [14]. In
well as a higher mean number of depressive episodes (irre-
terms of inpatient treatment, a lifetime prevalence of
spective of episode severity or duration). Furthermore,
24.1% for hospitalization owing to mental health prob-
higher levels of psychological and somatic comorbidity are
lems has been reported for PDD patients compared to
in line with international findings on higher comorbidity
12.1% for non-PDD patients [17]. Furthermore, the aver-
rates [12, 14, 15, 18, 20, 22] and somatic morbidity [14, 15]
age duration of past inpatient treatment is longer for
for chronic depression courses. The present outcomes of
PDD cases [53]. Patients with PDD also have higher
worse self-rated health and reduced health-related quality
rates of suicidal attempts and suicidal thoughts and are
of life for the MCS correspond to previous findings of re-
more likely to have a higher frequency of treatment ap-
duced psychological well-being and health-related quality
proaches in general and a longer disorder duration [22].
of life for individuals with chronic depression [12–14, 23].
Early depression onset seems a particular marker of a
Accordingly, chronic depression is associated with higher
more severe PDD course: 73% of our clinical PDD pa-
indirect economic costs: PDD cases showed a higher risk of
tients showed an early onset, whereas the proportion
experiencing limitation days owing to mental health prob-
was much lower among interview-defined PDD cases in
lems than non-chronic cases. Our findings of higher rates
DEGS1-MH (24.7%); and there was no significant differ-
of outpatient mental health care utilization and the trend
ence between PDD and NCMDD cases at a population
for a higher mean number of nights in hospital also indicate
level. Similarly, international findings are heterogeneous:
higher direct costs for the national economy and corres-
one meta-analysis found a significant relationship be-
pond to previous research findings [55].
tween early onset of depression and chronicity of the
disorder [54]. However, in a recent review of 17 studies
directly comparing age of onset in PDD vs. non-PDD Public health implications and future perspectives
cases, half the studies reported earlier onset for chronic Considering the growing frequency of depression and
vs. non-chronic depression whereas the other half re- health care costs in Germany owing to working days lost,
ported no difference [22]. early retirement, and health service provision [8, 56–58],
Recent reviews found that patients with PDD more our data strongly support the relevance of PDD as a spe-
often have psychiatric comorbidities than those with cific course of depressive disorders. As long-term PDD is
non-PDD, particularly personality disorders but also axis often associated with higher treatment resistance [59]
I and somatic comorbidities [22]. However, differences there is a chance that if an early and tailored treatment of
between PDD and NCMDD cases in DEGS1-MH have PDD and its specific psychopathological characteristicsNübel et al. BMC Psychiatry (2020) 20:58 Page 10 of 13 (e.g. CBASP) is carried out, a positive shift towards a more proved to be more effective than less specific psycho- positive course of the disease can be achieved. therapeutic treatments [75, 76]. However, self-reported utilization rates [see 52] corres- pond with reported international treatment gaps for men- Limitations tal disorders in general [3, 4, 6, 7]: most Germans with In interpreting the findings of this study some potential acute depression do not access mental health care. In limitations should be considered, such as the study de- addition, previous results indicate more frequent help- sign, response and reporting bias, and construct overlap. seeking with higher education [60]. The characteristics of The small number of PDD cases in both the clinical our clinical sample also suggest that in particular PDD sample and the DEGS1-MH sample may have reduced cases with lower educational levels do not seek help or re- the accuracy of the frequency and mean estimates. Thus, ceive (specialized) treatment: While international findings significant differences between the samples may not show that PDD is associated with low socioeconomic sta- have been detected using non-overlapping 95% CI. tus [61], PDD cases in our clinical sample had a signifi- Comparisons between PDD cases of DEGS1-MH and cantly higher education as compared to the general cases of the clinical sample are limited for several reasons. population (DEGS1-MH participants). This is important, Particularly severe and chronic depression cases may be as it raises the question of whether more educated pa- underrepresented in DEGS1-MH owing to the exclusion tients are more willing to participate in a depression inter- of institutionalized subjects, selective non-responses of less vention, or more likely to be informed about specific healthy individuals, and the inclusion of participants with treatment programs for PDD. If so, then PDD patients private health insurance as well as some longitudinal par- with lower education may be disadvantaged in this regard. ticipants (with a potentially greater probability of re- Moreover, findings from national health care data sug- participation among healthier persons) [33–35]. Moreover, gest that the validity of medical depression diagnoses are we found that clinical PDD patients had a higher educa- questionable, particularly in primary care [62], and that tional level than DEGS1-MH participants. This also limits improving treatment targeting [63, 64] and treatment the group comparison. However, it could indicate that pa- quality [19, 65–68] are desirable. In conclusion, these tients with PDD and a higher educational level have easier findings highlight the need for national public health ini- access to specified treatment programs. The comparison tiatives in Germany to reduce barriers to accessing men- between the DEGS1-MH and clinical samples is further tal health care services in general and in individuals with limited in terms of psychological comorbidity, owing to low education in particular, to strengthen awareness the use of different diagnostic tools (CIDI vs. SCID I). Fur- using targeted information campaigns, and to improve thermore, recall bias may have been more pronounced for the quality of medical recognition and specialized treat- DEGS1-MH cases, since PDD was defined on the basis of ment provision for depression and its different courses. lifetime information, whereas the clinical sample only in- There is thus a need to identify patients with PDD cluded patients diagnosed by PDD within the last 6 years. correctly and to tailor specific treatment strategies. Thus, recall bias may have led to an underestimation of de- Therefore, a focus on psychological characteristics [69, pression characteristics particularly among DEGS1-MH 70] is warranted, as the DSM-5 diagnosis of PDD is very cases, e.g. with regard to treatment resistance and a history likely a heterogeneous umbrella diagnosis. For example, of suicidality. Furthermore, recall bias, varying diagnostic different studies could differentiate PDD and non-PDD accuracy, and participants’ reporting bias may also have in terms of psychopathological features and social func- led to the underestimation of comorbidity and chronic de- tioning (e.g., cognitive and affective reactivity [69, 70] pression course during the lifetime in both surveys, par- and interpersonal behavior [71]. This is important for ticularly among male and older participants [64, 77]. the development of new treatment approaches as well as In DEGS1-MH, the small number of PDD and NCMDD for the empirical corroboration and refinement of exist- cases may also have led to low statistical power for detect- ing treatment attempts. For instance, CBASP was specif- ing the effects of depression course on health-related out- ically developed for the treatment of PDD [40]. CBASP comes. Furthermore, time lags between the core DEGS1 particularly considers psychopathological features of survey and its mental health supplement may have led to PDD such as an early onset due to childhood maltreat- an underestimation of associations between PDD/ ment and interpersonal withdrawal and avoidance. Evi- NCMDD and health-related correlates, as well as differing dence for the effectiveness of CBASP is encouraging (e.g. reference time frames for CIDI-based depression course [72]), especially in patients with childhood maltreatment during the lifetime and outcome variables (e.g., health ser- [73]. There is also evidence that the improvement of vice utilization during the past 12 months). However, con- interpersonal behavior through CBASP is associated struct overlap between depressive symptoms and the with symptom reduction, thus providing an important examined outcome measures (e.g., SF-36 and limitation treatment target for PDD [74]. In this regard, CBASP days) may have led to the overestimation of associations.
Nübel et al. BMC Psychiatry (2020) 20:58 Page 11 of 13
Conclusions Authors’ contributions
Finally, a chronic course of depression is challenging for JN and SM devised the main conceptual ideas. JN, SM, SK, and AG designed
the study. JN and HDL prepared and analyzed the data. JN and SK
both patients and practitioners. However, a knowledge interpreted the results and wrote a first draft of the manuscript. AG, SM, HDL,
gap remains regarding the lifetime characteristics and and CC contributed to the interpretation of results and critically revised the
correlates of chronic depression and the reliability of the article. All authors read and approved the final manuscript.
PDD concept itself. Funding
By combining clinical and epidemiological perspectives, This work was supported by the collaborative Focus Area DynAge project
our study permitted a comparison of standardized charac- (Freie Universität Berlin, Charité – Universitätsmedizin Berlin, German Institute
of Human Nutrition Potsdam-Rehbruecke (DIfE) and Robert Koch Institute),
teristics of PDD among clinical vs. non-clinical cases and the German Research Foundation (KO 5231/2–1), and by the German Minis-
therefore extends existing knowledge about PDD. Our try of Health (funding No. ZMV1–2516-FSB-703). DEGS1 and DEGS1-MH were
data suggest that the distinction between chronic and funded primarily by the German Ministry of Health (Bundesministerium für
Gesundheit, BMG, grant numbers for DEGS1-MH: 1368–1124 and 1501–
non-chronic depression proposed for DSM-5, in the form 54401). Supplementary funding for DEGS1-MH was provided by the Tech-
of PDD, is warranted. In particular, early onset depression, nische Universität Dresden, and by the Foundation for Mental Health (Stif-
attempted suicide, self-classification as persistent depres- tung Seelische Gesundheit) inaugurated by the German Association for
Psychiatry, Psychotherapy and Psychosomatics (Deutsche Gesellschaft für
sive course, and treatment resistance are suggested as Psychiatrie und Psychotherapie, Psychosomatik und Nervenheilkunde,
markers of more severe and chronic depression courses. DGPPN).
Furthermore, health-related correlates of PDD vs. non-
Availability of data and materials
chronic depression were compared at a population level.
Population-based data from the German health monitoring program that
Thus, the associated individual and economic disease bur- support the findings of this study are available from the Robert Koch
den was evaluated for the general population in Germany Institute (RKI) but restrictions apply to the availability of these data, which
were used under license for the current study and so are not publicly
for the first time. At a population level, chronic depression
available. The data set cannot be made publicly available because informed
is associated with a remarkably higher disease burden than consent from study participants did not cover public deposition of data.
non-chronic courses, indicating enormous direct and indir- However, a minimal data set is archived in the Health Monitoring Research
Data Centre at the RKI and can be accessed by all interested researchers. On-
ect costs of chronic depression for the national economy
site access to the data set is possible at the Secure Data Centre of the RKI’s
and emphasizing its public health relevancy. In conclusion, Health Monitoring Research Data Centre. Requests should be submitted to
these findings can inform the planning and targeting of the Health Monitoring Research Data Centre, Robert Koch Institute, Berlin,
Germany (Email: fdz@rki.de). The patient data from the clinical population are
prevention and health services. They highlight the need to
available from the Charité – Universitätsmedizin Berlin. The data set cannot
further reduce barriers to accessing mental health care, im- be made publicly available because informed consent from study partici-
prove awareness of different depression courses among pants did not cover public deposition of data.
health professionals, and implement specific treatment con-
Ethics approval and consent to participate
cepts for chronic depression. Research using the DEGS1-MH and the clinical sample at Charité – Universi-
tätsmedizin Berlin was conducted following the Federal and State Commis-
sioners for Data Protection guidelines and the ethical principles of the
Supplementary information Helsinki Declaration. Both surveys were approved by local ethics committees
Supplementary information accompanies this paper at https://doi.org/10. (Charité – Universitätsmedizin Berlin: No. EA2/047/08, No. EA4 140/13, Tech-
1186/s12888-020-2460-5. nische Universität Dresden: No. EK174062009). All participants provided writ-
ten informed consent prior to participation.
Additional file 1. Unadjusted effect estimates for health-related corre-
lates in cases of PDD vs. NCMDD (ref.) during the lifetime. Consent for publication
Not applicable.
Competing interests
Abbreviations
All authors declare that they have no competing interests.
95% CI: 95% confidence interval; CASMIN: Comparative Analysis of Social
Mobility in Industrial Nations; CBASP: Cognitive Behavioral Analysis System of
Author details
Psychotherapy; CIDI: Composite International Diagnostic Interview; 1
Department of Epidemiology and Health Monitoring, Unit 26 Mental Health,
DD: dysthymic disorder; DEGS1: German Health Interview and Examination
Robert Koch Institute, PO Box 650261, 13302 Berlin, Germany. 2Department
Survey for Adults; DEGS1-MH: Mental health module of DEGS1; DSM-IV-
of Psychiatry and Psychotherapy, Charité – Universitätsmedizin Berlin,
TR: Diagnostic and Statistical Manual of Mental Disorders, 4th Edition, Text
Campus Mitte, Charitéplatz 1, 10117 Berlin, Germany.
Revision; IRR: incidence rate ratio; MCS: mental component score;
MDD: major depressive disorder; NCMDD: non-chronic major depressive
Received: 13 August 2019 Accepted: 23 January 2020
disorder; OR: odds ratio; PCS: physical component score; PDD: persistent
depressive disorder; PHQ-9: Patient Health Questionnaire-9; RRR: relative risk
ratio; SCID I: Structured Clinical Interview for DSM-IV; SF-36: Short Form
Health Survey-36 References
1. GBD 2015 Disease and Injury Incidence and Prevalence Collaborators.
Global, regional, and national incidence, prevalence, and years lived with
Acknowledgments disability for 310 diseases and injuries, 1990–2015: a systematic analysis for
We thank Kim Ackermann for assisting during the follow-up assessment at the Global Burden of Disease Study 2015. Lancet. 2016;388:1545–602.
Charité – Universititäsmedizin Berlin. Furthermore, we thank Diane Williams, 2. GBD 2017 Disease and Injury Incidence and Prevalence Collaborators.
PhD, from Edanz Group (www.edanzediting.com/ac) for editing a draft of this Global, regional, and national incidence, prevalence, and years lived with
manuscript. disability for 354 diseases and injuries for 195 countries and territories,You can also read
